Not an issue - Need help to find a solution

[cloudbr34k84]

I finally found some ibeacon tags that i can use on dog colors :)

I just need some help in what settings i should use for dogs i guess??

They escaped once and we didnt know, so i want to avoid that happening.

I have Bluetooth proxys all over the house so essentially they should always be connected to a proxy. I have 2 proxies at the fromt of the hosue, in in the garage and in in the office.

What do you think would be the best way to set this up?

[agittins]

For the most part, it should "just work", and the less you experiment with the settings, perhaps the better!

Some things to keep in mind are:

• With the tag on their collar, at some times they are likely to be lying on top of it, which might prevent the tag appearing at all (because mammals are mostly water, and water is great at blocking 2.4GHz signals).
• Make sure you are using my suggested firmware settings for the proxies so that they forward their advertisements reliably and frequently - see the Wiki
• A bluetooth beacon can only tell you if the beacon is nearby. It can never tell you that a beacon is not nearby. Instead, we can only assume it's not nearby because we haven't seen it for a while. I know that sounds overly pedantic but it's super critical to understand. Think of it like a parole officer. They never know if you've scuppered off interstate, all they know is that they saw you yesterday, and expect to see you again tomorrow. The good thing about beacons is that they tend to advertise multiple times a second, and we have multiple proxies so that we are fairly confident that if the radio wave makes its way into free-air, we should hear it.

A simple workaround for the unreliable signal could be to install two beacons on each collar, on opposite sides - so that if puppo is sleeping on top of one of them, the other will have a clear view to the proxies. But it depends on the size of doggos and the cost of beacons if this is a sensible workaround for your situation.

To start with, I'd just try it out and see. You might find that even when sleeping on top of the tag there is enough signal, or you might get a lot of nuisance "drop-outs". There's no way to know ahead of time (and I only have cats).

As far as settings, sensors etc..

• Make max_radius very big, so that it has no effect. Something like 70m or something.
• devtrack_timeout is that main lever you have for controlling how long after the last signal the "Bermuda Device Tracker" sensors goes from "Home" to "Away". You want this number of seconds to be long enough that you don't get nuisance triggering, but short enough that you get notified they're "away" in a reasonable amount of time.
• Don't expect to be able to get a notification "as they pass through the gate" or "when they cross the fenceline". We just don't get that sort of definitive data from passively listening to radio signals.

Is there a particular path they always use to escape? If there's a "choke point" where they're not meant to be, you could put a proxy there, and alert if they are within x metres of that proxy at any point. This is unlikely, since it's usually gates or fences that they get through, but thought I'd mention it in case it applies.

If they have free access to outside, it gets a little harder because it's difficult to be sure if they're just hanging out or if they're just digging under the fenceline! If they're usually inside, you can create automations that check that they're within a certain distance of whichever area it is. Your automation could apply different distances based on the area, eg within 10m of the garage, or within 3m of the front door. What works will be highly dependent on your layout and needs.

What Bermuda gives you is sort of the "maximum" estimate. If you've calibrated your offsets and Bermuda says they are 3m away, then you can be pretty sure they are between 0m and 3m away. It's extremely unlikely they are 8m away, because the signal just couldn't have been that strong, but they could be closer because it's very likely the signal was partially blocked by something.

For the "Device Tracker Timeout" I'd normally go with 3 minutes or so (180s) as that's usually enough time to avoid nuisance triggering, but short enough that they're unlikely to have gotten far. I don't know if that's an acceptable figure for you, though.

You could create an automation that is a bit "twitchier" for some areas than others. So if the back yard fence is quite secure, but the front yard is where they're likely to escape, maybe check for shorter distances and also time-out more quickly if they were last seen close to the front door.

I know I haven't really given you any solid answers, only more questions! But with that in mind feel free to have a think about your situation and come back with any thoughts you have on how these things might apply, and if you want any help working through some options.

Obviously when I get trilateration implemented this sort of use-case should be much-improved, as you can then differentiate between a tag being between two proxies (ie, inside the house) or "on the far side" of a proxy.